Voice Biometrics Poised for Takeoff

Voice biometrics is expected to hit its stride in the next five years as its usage rises in the enterprise, among consumers, and in conjunction with other biometrics capabilities, according to recent IBM research.

Voice biometrics was a focal point of IBM's sixth annual "IBM 5 in 5," a list of innovations that, according to IBM, have the potential to change the way people work, live, and interact during the next five years.

IBM found that problems with voice biometrics, such as error rates in voice recognition, have been greatly reduced as the technology has developed. "We have seen great progress in the last three years," David Nahamoo, IBM fellow and chief technology officer for speech, says. "As performance gets better, we are seeing more adoption, and voice biometrics will hit the prime time, from small adoption to large, in the next five years."

Voice biometrics has seen usage climb in industries such as healthcare, thanks to the Health Information Portability and Accountability Act (HIPAA), which calls for strict identification methods. Use is also strong in the telecommunications space, and the financial industry is one of the fastest growing segments for voice biometrics adoption, especially when used with smartphones and tablets, and could be a game changer, according to Nahamoo.

"Imagine that you have a smartphone and you want to connect to your bank," Nahamoo says. "Initially, you have an ID and password, and you can do basic things. But if you want to transfer money or make a payment, these things require stronger authentication. When you request these applications, you would be engaged in the next level of verification that could involve speaker identification."

The voice biometrics process could be prone to piracy, Nahamoo points out, if someone records your voice saying your password. However, advanced voice biometrics could employ a more sophisticated system, such as asking the speaker unrelated questions, making it difficult for someone to predict what the next prompt would be.

"Random questions are recommended so that they complement pure biometrics with knowledge-based answers to reduce false acceptance rates," Nahamoo says.

With the proliferation of smartphones and tablets, Nahamoo says voice biometrics technology is also now becoming integrated into other biometrics modalities. "For the first time, we are seeing real devices that mix multiple biometrics—voice, facial recognition, and gesture verification—that go a long way in creating access that is natural and at the same time provides security that we all need," he says.

"It will take some time to work out the usability aspect and integration into different business processes, evaluation and testing, but usability is key," Nahamoo says.